We’re on the verge of a paradigm shift in the way geospatial data is collected for commercial, scientific, and recreational purposes. Unmanned aerial systems (hereafter called drones) have the ability to collect data with higher speed, accuracy, and resolution than conventional methods, and their uses and applications will only continue to accelerate.

As a co-founder of Umap Technologies, I saw what kind of an advantage drones possessed when compared to more traditional data capture methods, and wanted to introduce those advantages to engineering, surveying, and construction professionals with our image processing platform. However, being able to effectively leverage these advantages is about far more than a particular solution. So much of it can be contingent on something far more basic when it comes to how they’re handling the data captured by a drone.



Drone Advantages

I probably don’t have to convince anyone of the advantages than drones can provide to projects of all sizes and scopes, but it always makes sense to be as specific as possible with this technology.

Imagine that you’re acquiring 3D data for a construction worksite. Using conventional methods, such as ground based surveying measurements (e.g. theodolite/EDM, GNSS, etc.) it may take a significant amount of time to collect as-built data from this site due to it’s size. Additionally, you may not be able to effectively locate building corners, towers, or any other structure that are inaccessible to ground based measurement.

That’s where drones can and do make a difference. Using photogrammetry in conjunction with drones, you can perform data collection on a whole site in a matter of minutes, and the resulting data is far denser (millions of 3D points) than even the most dedicated surveying technician is capable of collecting.

That being said, there are limitations. If drone users are producing high-resolution 3D models from imagery (e.g. multi-view stereo photogrammetric methods), the reconstructions are still limited by what the camera can see.  For example, we cannot “see” through vegetation to the ground, whereas an active sensor such as LiDAR may in fact be able to pick up limited ground information (i.e. not just the first return).

Limitations are what they are, but it’s been incredible to see so many different amounts and types of data come through our pipeline. While this is probably just a spattering of what’s out there, some of those different data types include:

Passive Sensors (Imagery/Video)

  • Visible (RGB)
  • Near Infrared (NIR)
  • Thermal (Far Infrared)
  • Multispectral
  • Hyperspectral


Active Sensors

  • LiDAR
  • Radar

What’s really amazing to think about is the fact that these are just the different sensor types. The actual applications are almost too numerous to count, and all of it seems to change and develop on a daily basis.

What’s important to remember is that drone-based data acquisition is another “tool in the toolbox” for many different industries professions. It doesn’t solve every problem that exists for 3D measurement, but it does solve many of them more efficiently than was previously possible, and we’ve even come a long way in terms of those logistics. Need a drone with a metric camera? Done. Need a thermal imaging sensor that can be easily integrated on your drone? No problem. There are still times when there isn’t an exact match between the drone platform and the payload, but those instances seem to be occurring less and less as time goes on.

This “capture” phase of the workflow is often the most exciting and the one people talk about the most, but where is the data that’s being captured by a drone going? How is it being processed and stored? These are the sorts of questions that are becoming as pervasive as they are important.


What do you do with all this data?

Questions around what can and should be done with all this data are ones that professionals of all sizes and types are struggling with, and part of that is because we really don’t have a set of standards around these issues. It represents a big problem that’s starting to become more apparent as the drone industry matures and companies scale up in their capabilities.

What I can tell you from experience is that successes and challenges in this area all depend on the individual and the level of sophistication they seek to provide to their customer. Some companies still store their data on portable hard drives that reside in the office closet collecting dust for 365 days a year. Others prefer to provide a cloud-based data storage solution as a service to their end customer. We have a processing platform (Umap) that includes lifetime data storage. Suffice to say, there are a number of options out there, and they range in terms of their limitations, price and capabilities. Finding the right fit is more about your project and process than anything else though.

Regardless of the solution you move forward with, drone-based data collection is “Big Data”.  If you’re flying 3-5 times/week and averaging 500 images per flight, you’re amassing tens of gigabytes each week and you need to make a plan about what’s going to happen with all this info, and specific choices in this area depend on what you need to do.

Those choices will impact the present and future of the project. Most operators prefer to keep everything, provided the cost is within reason. We’ve all been burned enough times to know that being able to resurrect old projects with minimal effort can be invaluable in many cases.



When should you be making decisions about your data?

As the King in Alice in Wonderland said, it’s very important to “begin at the beginning”, and starting with a plan or course of action around your data is arguably more important than any other decision you can make.

In many cases, stakeholders may not understand what the operator is capable of producing, or the formats that are produced, so it’s always worth spending the extra time up front to inform them of the exact deliverable, its accuracy, and any weaknesses. They may request “3D data”, but not actually be capable of using 3D point cloud or 3D mesh data effectively in their existing workflow. After some discussion, an operator may be surprised to find a stakeholder really defines “3D Data” as a 2D map with contours overlaid. Some clients are incredibly knowledgeable about their workflows and what deliverables they need. Other clients are just dipping their toes into 3D data, or worse yet, they have strong opinions on what they think they need, but have little understanding of how and where it is needed. Just a reminder, whoever is paying the bill is always right.

Data is where all of the value lies so a smart service provider will only provide the data the stakeholders are paying for, and in some cases, that’s the only data they’ll be able to handle. Service providers can increase their billings when they educate their customers on the value and applications of those extra data products, which means it’s just as important for people on both sides of the table to understand the value.

It’s quite clear to me that the greatest challenge facing the industry is the overall confusion around these topics. We’re dealing with a wide variance of solution quality, lots of brand new solutions providers, and a lot of Socratic “expects”. The best thing service providers can do is to take their education in this exciting space seriously and engage with customers, clients and stakeholders to help make them aware of the opportunities that are available.


Data Challenges and Solutions

Remote areas are challenging. If you can’t push data out to the cloud or back to the home office easily, you’re in a pickle. That being said, there are commercial solutions available that essentially provide a mobilized processing environment that can tackle large remote projects that are a long ways from the grid.

This is a perfect example of why it’s important for the operator to make sure that all of the project specifics are laid out in detail before data collection. What is the turnaround time on the deliverable/s?  What are the accuracy requirements? Who is the end user of the deliverables?

We’ve dealt with issues that are far more basic though. We know a number of people that hand deliver data to customers with flash drives. Of course, if the client specifically requests that delivery method, then by all means do what’s best for the project. However, most are starting to appreciate the value that a data storage and access service brings to the table.

There are already a number of storage solutions out there that allow operators to transfer data to clients efficiently without needing to be a software engineer. Dropbox, Box, Drive, etc, The weakness of these generic storage solutions is a lack of intimate understanding of the data and how to exploit it. You cannot view a point cloud in Dropbox, or painlessly overlay an orthoimage in Google Earth, but you certainly can in a data-aware solution specifically designed for processing and storing 3D data.



Data and Deliverables

Being able to effectively and concisely identify what a deliverable for a project is going to look like can eliminate a number of headaches for everyone, but that means expectations around the needs and capabilities of a process and system  have to be effectively communicated. It’s important to communicate as often as possible, even if it means admitting a mistake, even if you’re asking simple questions. It’s always better to ask a dumb question than put yourself in a dumb situation.

With the amount of info that a drone can gather it’s easy to get overwhelmed, but a concise plan and solution can make it just as simple to stay on top of everything. There are a number of image processing platforms out there, and you should look around to figure out which one is going to be best suited to your needs.

If you have the budget to spend on a more expensive solution, is it in your best interest to go with that option or use that money for something else? Can you get away with putting everything on DropBox, or will choosing a solution like Tru.VU make a real difference? These are decisions you need to make, but just asking the questions is an important first step.